Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Efficient Indexing for Object Recognition Using Large Networks \Lambda Mark R. Stevens Charles W. Anderson J. Ross Beveridge
 

Summary: Efficient Indexing for Object Recognition Using Large Networks \Lambda
Mark R. Stevens Charles W. Anderson J. Ross Beveridge
Department of Computer Science
Colorado State University
Fort Collins, CO 80523
fstevensm,anderson,rossg@cs.colostate.edu
Abstract
Template matching is an effective means of locating vehicles in outdoor scenes, but it tends to be
a computationally expensive. To reduce processing time, we use large neural networks to predict, or
index, a small subset of templates that are likely to match each window in an image. Results on actual
LADAR range images show that limiting the templates to those selected by the neural networks reduces
the computation time by a factor of 5 without sacrificing the accuracy of the results.
1 Introduction
Neural networks are often used to extract complex, nonlinear relationships among the variables of a set
of data. However, in this article we use the nonlinear mapping capability of neural networks to reduce
the computation associated with the essentially linear procedure of template matching for automatic target
recognition (ATR). Template matching for vehicle identification requires the application of numerous tem­
plates to rectangular windows of an image [12, 8, 5, 6]. Each template corresponds to a particular type of
vehicle at a particular orientation. The windows are small, rectangular subsets of pixels just large enough
to contain each vehicle and be within the image bounds. After all templates are applied to an image, the

  

Source: Anderson, Charles W. - Department of Computer Science, Colorado State University

 

Collections: Computer Technologies and Information Sciences